- Email: [email protected]
Changes in Hemoglobin Level Distribution in US Dialysis Patients From June 2006 to November 2008
Changes in Hemoglobin Level Distribution in US Dialysis Patients From June 2006 to November 2008 David M. Spiegel, MD,1 Irfan Khan, PhD,2 Mahesh Krishnan, MD, MPH,2 and Tracy J. Mayne, PhD2 Background: Erythropoiesis-stimulating agents (ESAs) have had a positive effect on anemia treatment in dialysis patients. However, several events in recent years, including new clinical study results, ESA product label revisions, and coverage and reimbursement policy changes, have had an impact on ESA dosing patterns and consequently on hemoglobin (Hb) distribution characteristics in this patient population. Study Design: Retrospective observational study using patient-level data from ⬃87% of dialysis centers in the United States. Setting & Participants: Dialysis patients who were receiving outpatient care at dialysis facilities during June 2006-November 2008 were included in this study. Predictor: Recent events affecting ESA treatment practice patterns in US dialysis patients. Outcomes & Measurements: Hb level distribution. Results: Mean Hb level decreased by 0.37 g/dL during the indicated period. Additionally, standard deviation (SD) of the Hb level distribution decreased by 0.14 g/dL and skewness increased by ⫺0.10. Hb measurements in specific ranges changed as follows: ⬎12 g/dL, decreased by 11.3 percentage points;10-12 g/dL, increased by 9.4 percentage points; and ⬍10 g/dL, increased by 1.9 percentage points. The percentage of patients with Hb level ⬎13 g/dL for ⱖ3 months decreased by 2.9 percentage points. Limitations: Potential bias in dialysis center selection and lack of information for patient characteristics. Conclusions: Recent events affecting ESA use in dialysis patients have had the desired effect of increasing the proportion of Hb measurements within the US Food and Drug Administration recommended target range of 10-12 g/dL and decreasing the proportion of Hb measurements ⬎12 g/dL. However, the proportion of Hb measurements ⬍10 g/dL also has increased. Benefits of a decrease in Hb measurements in the ⬎12 g/dL range need to be considered, together with risks of having low Hb levels. Am J Kidney Dis 55:113-120. © 2009 by the National Kidney Foundation, Inc. INDEX WORDS: Hemoglobin (Hb) distribution; dialysis; erythropoiesis-stimulating agents; Hb standard deviation.
T
he practice of anemia management in dialysis patients has evolved significantly since the US Food and Drug Administration (FDA) approved the first erythropoiesis-stimulating agent (ESA), epoetin alfa, in 1989. The goal of anemia management since then has been a decrease in the proportion of patients with a less-than-target hemoglobin (Hb) level, driven by results of observational studies indicating an association between low Hb level and increased morbidity and mortality risk,1-7 as well as the established clinical benefit shown in randomized trials of transfusion avoidance8,9 and enhanced physical function.10-14 This evidence also was reflected in the 1997 Kidney Disease Outcomes Quality Initiative (KDOQI) clinical practice guidelines from the National Kidney Foundation (NKF) for anemia management in dialysis patients that recommend a Hb target range of 11-12 g/dL.15 As a result, the mean Hb level of the US dialysis
patient population increased from 9.7 g/dL in 1991 to nearly 12.0 g/dL in 2006.16 In late 2006 and 2007, multiple events took place that fundamentally changed ESA dosing practices in dialysis patients. First, new clinical trial data emerged showing either increased adverse events or no benefits in targeting higher Hb levels. The CHOIR (Correction of Hemoglobin
From the 1Health Sciences Center, University of Colorado, Denver, CO; and 2Department of Health Economics, Amgen Inc, Thousand Oaks, CA. Received May 22, 2009. Accepted in revised form September 24, 2009. Originally published online as doi:10.1053/j. ajkd.2009.09.024 on November 23, 2009. Address correspondence to David M. Spiegel, MD, 4545 E 9th Ave, Ste 160, Denver, CO 80220. E-mail: david. [email protected] © 2009 by the National Kidney Foundation, Inc. 0272-6386/09/5501-0016$36.00/0 doi:10.1053/j.ajkd.2009.09.024
American Journal of Kidney Diseases, Vol 55, No 1 (January), 2010: pp 113-120
113
114
Spiegel et al
and Outcomes in Renal Insufficiency)17 and CREATE (Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta)18 studies used ESAs to increase Hb levels in patients with chronic kidney disease not on dialysis therapy. In the CHOIR study, patients targeted to a high Hb level (13.5 g/dL) had statistically significantly more composite end point events than those targeted to lower Hb levels (11.3 g/dL). In CREATE, no benefits were found in targeting patients to a Hb level of 13.0-15.0 g/dL compared with patients targeted to a Hb level of 10.5-11.5 g/dL. Both of these studies supported what had been shown previously in the Normal Hematocrit Cardiac Trial (NHCT).19 As a result of these clinical trial data, the FDA issued a boxed warning and other product labeling changes for ESA treatment in patients with chronic kidney disease (both on and not on dialysis therapy) in March 2007. The August 2008 update of the product label recommended a Hb level range of 10-12 g/dL and included specific dose adjustment guidelines to achieve these targets, as well as guidance for dosing in hyporesponsive patients.20,21 In addition, the NKF KDOQI panel convened in March 2007 to consider findings from the new studies and issued a practice guideline update that recommended a target Hb level of 11-12 g/dL, rather than a target ⱖ11 g/dL.22 Finally, in July 2007, the Centers for Medicare & Medicaid Services (CMS) communicated changes to its existing ESA monitoring policy. In addition to encouraging downward titration in response to high Hb levels, the revised policy, effective as of January 1, 2008, decreased the maximum monthly epoetin alfa and darbepoetin alfa doses that would be reimbursed to 400,000 U/mo and 1,200 g/mo, respectively. The revised policy also decreased ESA coverage by 50% if Hb levels were ⬎13 g/dL during the current month and prior 2 months.23 The purpose of this study is to assess Hb level distribution characteristics of the dialysis patient population in the United States during the time frame when these events occurred.
METHODS A retrospective observational database was constructed from electronic patient-level data submitted by a large number (⬃4,800) of dialysis centers in the United States. These
data were obtained and purchased as part of product purchase and sales agreements maintained between Amgen Inc and its customers. Dialysis units transferred the data electronically to Thomson Reuters Healthcare Inc, which took steps to ensure proper data integration, quality assurance, and compliance with the federal Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule. The data set then was certified by an outside consultant as statistically deidentified under the HIPAA privacy rule for use within Amgen Inc; therefore, no HIPAA waiver was sought or required. The corresponding author (D.M.S.) was not privy to individually identifiable patient-level data (reviewing instead composite data analyses and graphs) and therefore no institutional review board approval was sought for his work on the data set. We classified dialysis facilities as large dialysis organizations (defined as a chain ⬎500 affiliated units), small dialysis organizations, and hospital-based dialysis centers. Of nearly 5,500 dialysis centers currently active in the United States, ⬃60% are large dialysis organizations, 30% are small dialysis organizations, and 10% are hospital-based dialysis centers. The present study database represents 100% of facilities from the large dialysis organization segment (⬃3,300) and 90% of facilities from the small dialysis organization segment (⬃1,500). Overall, nearly 87% of dialysis centers in the United States (⬃4,800) are represented. The primary reason for exclusion of facilities (ie, 100% of hospital-based dialysis centers and 10% of small dialysis organizations) was practical limitations in accessing their data because of logistical or technical hurdles. Dialysis centers that are captured in the database submitted data for all patients receiving care within that unit, and no exclusion criteria were applied. Thus, patients with all modality types, including hemodialysis and peritoneal dialysis, and patients with all coverage types, including commercial and Medicare, were represented. The database contains laboratory measurements, including all Hb measurements and intravenous medications administered by dialysis centers. Although laboratory measurements were available for all patients in the database, intravenous medications were available for only patients in large dialysis organization dialysis facilities. The primary variables used in the present study were Hb measurements. All available Hb measurements for each patient (including multiple measurements per month) were used in the analysis. The study time frame was June 2006-November 2008. This time frame was selected because November 2008 was the latest month when data were extracted for this study, and a 30 month history was judged to be sufficient for qualitative and quantitative assessment of changes due to events following publication of the CHOIR17 and CREATE18 studies in late 2006. In addition, June 2006 was judged to be a stable “baseline” point relative to which future changes can be assessed because anemia management practice patterns during this period were relatively stable and uninfluenced by future ESA safety-related events.16 The following sets of variables were calculated for each month during the study time frame for the cross-section of active patients: (1) mean Hb level; (2) standard deviation (SD) of Hb distribution; (3) skewness of Hb distribution; (4) continuous Hb distribution curve covering 0.1 g/dL intervals; (5)
Hemoglobin Distribution Changes in Dialysis percentage of Hb measurements in the following 3 ranges: ⬍10, 10-12, and ⬎12 g/dL; and (6) percentage of patients with Hb level ⬎13 g/dL for ⱖ3 months. All these calculations, except for item 6, were based on all available patient-level Hb measurements during a given month; multiple Hb values in a month were not averaged. This procedure was adopted because we were interested in describing the raw Hb distribution curve that represented total variation in the data, including within- and betweenpatient variation. Although within-patient variation during a single month was very small, averaging would have masked the information represented in this source of variation and the marginal contribution from outliers. We also performed calculations in which patient-level Hb values were averaged during a month, and the results appeared to be nearly identical to those reported in this study. Skewness calculation in item 3 represents a measure of the departure from normality for the Hb distribution curve. Skewness is zero for a normally distributed curve and nonzero if the distribution has a “tapering” characteristic toward either end. Positive skewness indicates tapering to the right, whereas negative skewness denotes tapering toward the left. The evaluation of changes in skewness can inform how the reduction in the width of Hb distribution curve has been accomplished. Calculation of the percentage of patients with a Hb level ⬎13 g/dL for ⱖ3 months (item 6) was performed considering only the last recorded Hb value for individual patients during the month. This procedure was adopted to be consistent with the manner in which the CMS calculates this information for the purpose of compliance assessment with ESA monitoring policy. The continuous Hb distribution curve (item 4) represented a smooth approximation of Hb measurement density within 0.1 g/dL intervals. This smooth approximation was necessary to avoid unnecessary visual attention to density fluctuations driven by either low prevalence buckets or rounding of reported Hb measurements to
Figure 1. Time trend of mean hemoglobin (Hb) levels. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
115 the nearest decimal by some laboratories and was obtained using PROC TPSPLINE in SAS, (SAS Institute Inc, www.sas. com) a standard smoothing tool for this purpose (comparisons with unsmoothed data also confirmed the appropriateness of this method). All analyses were conducted using SAS, version 9.1.3. P values and confidence intervals for changes discussed in results are not reported because the small P values and narrow confidence intervals (the result of a very large patient number) were not informative.
RESULTS Approximately 743,000 unique patients were available in the analysis data set during the study time frame covering June 2006-November 2008 (study period prevalence). Mean follow-up for these patients during the study time frame was 12.3 months (5th-95th percentiles, 0.5-29.5). The cohort of patients alive during any month represented the sum of those who survived from the previous month and those who represented new patients added during the month. The mean number of unique patients per month (ie, the monthly prevalence) was ⬃309,000. These patients represented ⬃734,000 unique Hb measurements each month. Mean Hb measurement frequency per patient per month was ⬃2.4 ⫾ 3.5 (SD). Mean population Hb level decreased from 12.08 ⫾ 1.48 g/dL in June 2006 to 11.71 ⫾ 1.35 g/dL in November 2008, a decrease of 0.37 g/dL (Fig 1). The periods of sharpest decrease in mean population Hb level coincided with the anticipa-
116
Spiegel et al
Figure 2. Time trend of the standard deviation (SD) of hemoglobin (Hb) distribution. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
tion and occurrence of the following events: revision of the ESA label by the FDA in March 2007, updates to the NKF KDOQI clinical practice guidelines in March 2007, and implementation of the revised ESA monitoring policy by the CMS in January 2008. The highest mean population Hb level during the study time frame was 12.14 ⫾ 1.50 g/dL in January 2007, whereas the lowest mean population Hb level was 11.71 ⫾ 1.35 g/dL in November 2008, a difference of 0.44 g/dL. In a typical cross-sectional sample of dialysis patients at any point in time, Hb values cover a wide range around the mean, and the resulting distribution is approximately normal.24 The SD has been used as a measure of the width of the Hb distribution curve. SD was calculated for each month to assess potential changes in the width of the Hb distribution curve over time. SD decreased from 1.48 g/dL in June 2006 to 1.35 g/dL in November 2008 (decrease of 0.14 g/dL; Fig 2). Figure 3 shows the skewness of the Hb distribution curve over time. The Hb distribution curve showed a departure from normality in terms of skewness. A general negative skew for the Hb distribution curve indicates that the historical Hb distribution curve probably has been skewed toward the left, indicating that significantly low Hb values relative to the peak were more common compared with significantly high
Hb values. As the SD of Hb distribution curve has decreased, the magnitude of negative skewness has increased simultaneously from ⫺0.11 in June 2006 to ⫺0.21 in November 2008, an absolute increase of 0.10. Together, these data indicate that the reduction in width of the Hb distribution curve has resulted primarily from migration of patients from high Hb ranges toward the middle. This observation also is readily apparent in Fig 4, which shows changes in continuous Hb distribution curve over time. A detailed examination of Fig 4 indicates that the area under the curve between 10 and 12 g/dL has systematically increased, primarily because of a net patient migration from high Hb ranges. In addition, the entire distribution curve appears to have shifted toward lower Hb ranges. Figure 5 provides a more granular look at the dynamics in specific Hb ranges. The percentage of Hb measurements in the ⬎12 g/dL range decreased from 51.3% in June 2006 to 40.1% in November 2008 (a decrease of 11.3 percentage points). However, the percentage of Hb measurements in the FDA-recommended target range of 10 12 g/dL increased from 41.3% in June 2006 to 50.6% in November 2008 (an increase of 9.4 percentage points), and the percentage of Hb measurements in the ⬍10 g/dL range also increased by 1.9 percentage points from 7.4% in June 2006 to 9.3% in November 2008.
Hemoglobin Distribution Changes in Dialysis
117
Figure 3. Time trend of the skewness of hemoglobin (Hb) distribution. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
Figure 6 shows that the proportion of patients with Hb levels persistently ⬎13 g/dL for ⱖ3 months decreased from 5.1% in June 2006 to 2.2% in November 2008 (a decrease of 2.9 percentage points). ESA-dosing behavior was examined for patients with a Hb level persistently ⬎13 g/dL for at least 3 months in November 2008 who also received care in large dialysis organizations. Approximately 64% of these patients received zero ESA dose during November 2008. That is, these patients naturally remained
in the Hb ⬎13 g/dL range without requiring further administration of an ESA.
DISCUSSION Changes implemented by the FDA and CMS in light of safety concerns regarding ESA treatment in dialysis patients have had a profound impact on anemia management practice patterns in the United States. A population-level Hb distribution curve that would be aligned best with the intent of these changes would maximize the
Figure 4. Continuous hemoglobin (Hb) distribution curve over time.
118
Spiegel et al
Figure 5. Percentage of hemoglobin (Hb) measurements in specific ranges. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
number of patients in the target range of 10-12 g/dL, minimize the number of patients in the ⬍10 and ⬎12 g/dL ranges, and minimize the number of patients who are serially exposed to ⬎13 g/dL. To accomplish the desired result of maximizing patients in the target range, one could affect the mean population Hb level, SD, or skewness characteristics of the Hb distribution curve. The Hb distribution curve described in the present study represents a “snapshot,” or a crosssectional point-in-time view. However, when in-
Figure 6. Percentage of patients with hemoglobin (Hb) ⬎13 g/dL for ⱖ3 months. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
dividual patients are followed up over time, they are seen to oscillate between low, intermediate, and high Hb ranges, and extremely few patients are successfully managed in a narrow range over time.24 It is not clear whether these oscillations represent an intrinsic clinical characteristic of patients, transient intercurrent events, or are influenced by anemia management practice patterns. The nature of the snapshot Hb distribution curve thus is influenced largely by the nature of the patient migration pattern between low, intermedi-
Hemoglobin Distribution Changes in Dialysis
ate, and high Hb ranges. The information regarding changes in mean, SD, and skewness of the Hb distribution curve from this study can be used to generate useful hypotheses about the types of patient migration patterns that are consistent with these changes. Although the decrease in mean population Hb level presumably represents decreased ESA dosing and lower point targets, both the decrease in the SD and increase in the negative skewness could be explained by a preferential patient migration from high Hb ranges toward the target range, accompanied by some patient migration from target range to below target. Although we do not yet understand the exact changes in the nature of ESA dosing algorithms that might have led to the observed changes in the SD and skewness of the Hb distribution curve, it is an area worthy of further investigation. One possible explanation for the observed decrease in the SD of the Hb distribution curve, as shown in Fig 2, may be changes in specific Hb thresholds that trigger a downward dose adjustment, as well as changes in the magnitude and frequency of these downward dose adjustments in response to the factors described in this study. Other factors, such as changes and/or integration of iron and ESA dosing protocols, also may have contributed to the observed changes in the Hb distribution curve.25 Based on current guidelines, ESA dosing protocols should strive to maximize the number of patients with Hb levels in the 10-12 g/dL range while minimizing the proportion of patients in the high-Hb range. However, this should not be at the expense of an increase in the proportion of patients in the low-Hb range. To achieve this level of control, anemia management practice patterns that allow for timely changes in Hb levels through more frequent Hb monitoring and more frequent dose adjustments may be preferred over anemia management practice patterns based on infrequent Hb measurements requiring potentially large dose changes. Additional factors, such as the nature of ESA dose-holding practices, also may influence the nature of Hb distribution. For example, a recent study evaluating individual patient-level data in a single center suggested that a principal cause of Hb level decreases to ⬍11 g/dL was withholding of ESAs for a Hb level ⬎13 g/dL.26 However, not all published literature support this conclusion.27 If
119
ESA doses are withheld to manage high Hb level excursions, it seems prudent to monitor Hb levels frequently so that treatment is reinitiated before Hb levels decrease too much. More studies evaluating population-level data are needed to understand if a more optimal approach toward ESA dosing algorithms can be developed. Some important limitations of our study need to be considered. Although the sample size was relatively large, incorporating data from nearly 87% of dialysis centers in the United States, patients in the remaining 13% of dialysis centers may represent a systemic bias because excluded facilities represented 100% of the hospital-based dialysis center facilities and 10% of the small dialysis organization facilities. Although the excluded hospital-based dialysis center segment could have different patient and anemia management characteristics from the facilities studied, the 10% excluded small dialysis organization facilities also could have characteristics that are different compared with the rest of the small dialysis organization segment because the excluded small dialysis organization facilities were primarily those that did not have a fully functional electronic system for patient-level data collection. Another limitation of our study was that we did not have demographic or comorbidity information for patients in the analysis database and thus were unable to summarize patient clinical characteristics. However, for the facility segments under consideration in this study, these characteristics have been well documented and reported in several other sources.5,6,16,24 In addition, we consider it to be extremely unlikely for potential changes in patient characteristics to be an explanation for the dynamics reported in this study because patient characteristics change over a much longer time horizon and anemia management practices have a far greater and immediate impact on the nature of Hb level distribution in this patient population. Another limitation of the study is that although Hb data from outpatient dialysis sessions have been represented, data from inpatient hospitalization records were not included. However, because only ⬃4% of dialysis sessions per patient-year occur in the inpatient setting,16 any potential changes in practice patterns during hospitalizations are very unlikely to explain the population-level results discussed in this study.
120
Spiegel et al
Despite these limitations, this study offers important findings regarding changes in Hb level distribution characteristics for a large number of patients. How these changes will ultimately affect patient outcomes in the longer term is unclear and is a topic worthy of further investigation.
ACKNOWLEDGEMENTS The authors thank Santosh Sastry, MS, for help with data access and SAS programming and Yeshi Mikyas, PhD, for editorial support. Support: Information on funding sources is listed in the financial disclosure. Financial Disclosure: This study was supported by Amgen Inc, which markets ESAs. Dr Spiegel has received consulting income and grant support from Amgen Inc and has served on advisory boards and speakers bureaus for Amgen Inc. Drs Khan, Krishnan, and Mayne are employees of Amgen Inc.
REFERENCES 1. Collins AJ, Li S, St. Peter W, et al. Death, hospitalization, and economic associations among incident hemodialysis patients with hematocrit values of 36 to 39%. J Am Soc Nephrol. 2001;12:2465-2473. 2. Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE. The impact of anemia on cardiomyopathy, morbidity, and mortality in end-stage renal disease. Am J Kidney Dis. 1996;28:53-61. 3. Lowrie EG, Huang WH, Lew NL, Liu Y. The relative contribution of measured variables to death risk among hemodialysis patients. In: Hingman FE, ed. Death on Hemodialysis: Preventable or Inevitable. Boston, MA: Kluwer Academic Publishers; 1994:121-141. 4. Ma JZ, Ebben J, Xia H, Collins AJ. Hematocrit level and associated mortality in hemodialysis patients. J Am Soc Nephrol. 1999;10:610-619. 5. Ofsthun N, Labrecque J, Lacson E, Keen M, Lazarus JM. The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients. Kidney Int. 2003;63:1908-1914. 6. Regidor DL, Kopple JD, Kovesdy CP, et al. Associations between changes in hemoglobin and administered erythropoiesis-stimulating agent and survival in hemodialysis patients. J Am Soc Nephrol. 2006;17:1181-1191. 7. Roberts TL, Foley RN, Weinhandl ED, Gilbertson DT, Collins AJ. Anaemia and mortality in haemodialysis patients: interaction of propensity score for predicted anaemia and actual haemoglobin levels. Nephrol Dial Transplant. 2006;21:1652-1662. 8. Eschbach JW, Abdulhadi MH, Browne JK, et al. Recombinant human erythropoietin in anemic patients with end-stage renal disease. Results of a phase III multicenter clinical trial. Ann Intern Med. 1989;111:992-1000. 9. Eschbach JW, Egrie JC, Downing MR, Browne JK, Adamson JW. Correction of the anemia of end-stage renal disease with recombinant human erythropoietin. Results of a combined phase I and II clinical trial. N Engl J Med. 1987;316:73-78.
10. Canadian Erythropoietin Study Group. Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving haemodialysis. BMJ. 1990;300:573-578. 11. Beusterien KM, Nissenson AR, Port FK, Kelly M, Steinwald B, Ware JE Jr. The effects of recombinant human erythropoietin on functional health and well-being in chronic dialysis patients. J Am Soc Nephrol. 1996;7:763-773. 12. Evans RW, Manninen DL, Garrison LP Jr, et al. The quality of life of patients with end-stage renal disease. N Engl J Med. 1985;312:553-559. 13. Evans RW, Rader B, Manninen DL. The quality of life of hemodialysis recipients treated with recombinant human erythropoietin. Cooperative Multicenter EPO Clinical Trial Group. JAMA. 1990;263:825-830. 14. Laupacis A. Changes in quality of life and functional capacity in hemodialysis patients treated with recombinant human erythropoietin. The Canadian Erythropoietin Study Group. Semin Nephrol. 1990;10:11-19. 15. National Kidney Foundation. DOQI Clinical Practice Guidelines for the Treatment of Anemia of Chronic Renal Failure. Am J Kidney Dis. 1997;30(suppl 3):S192-240. 16. US Renal Data System. USRDS 2008 Annual Data Report. Bethesda, MD: The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2008. 17. Singh AK, Szczech L, Tang KL, et al. Correction of anemia with epoetin alfa in chronic kidney disease. N Engl J Med. 2006;355:2085-2098. 18. Drueke TB, Locatelli F, Clyne N, et al. Normalization of hemoglobin level in patients with chronic kidney disease and anemia. N Engl J Med. 2006;355:2071-2084. 19. Besarab A, Bolton WK, Browne JK, et al. The effects of normal as compared with low hematocrit values in patients with cardiac disease who are receiving hemodialysis and epoetin. N Engl J Med. 1998;339:584-590. 20. Epogen (epoetin alfa) [prescribing information]. Thousand Oaks, CA: Amgen Inc; 2008. 21. Aranesp (darbepoetin alfa) [prescribing information]. Thousand Oaks, CA: Amgen Inc; 2008. 22. National Kidney Foundation. KDOQI Clinical Practice Guideline and Clinical Practice Recommendations for Anemia in Chronic Kidney Disease: 2007 update of hemoglobin target. Am J Kidney Dis. 2007;50:471-530. 23. Centers for Medicare & Medicaid Services. CMS Manual System. Pub 100-04 Medicare Claims Processing. Transmittal 1307. http://www.cms.hhs.gov/Transmittals/ Downloads/R1307CP.pdf. Accessed January 15, 2009. 24. Lacson E Jr, Ofsthun N, Lazarus JM. Effect of variability in anemia management on hemoglobin outcomes in ESRD. Am J Kidney Dis. 2003;41:111-124. 25. Chan K, Moran J, Hlatky M, Lafayette R. Protocol adherence and the ability to achieve target haemoglobin levels in haemodialysis patients. Nephrol Dial Transplant. 2009;24:1956-1962. 26. Spiegel DM, Gitlin M, Mayne T. Factors affecting anemia management in hemodialysis patients: a singlecenter experience. Hemodial Int. 2008;12:336-341. 27. Weiner DE, Miskulin DC, Seefeld K, et al. Reducing versus discontinuing erythropoietin at high hemoglobin levels. J Am Soc Nephrol. 2007;18:3184-3191.
T
he practice of anemia management in dialysis patients has evolved significantly since the US Food and Drug Administration (FDA) approved the first erythropoiesis-stimulating agent (ESA), epoetin alfa, in 1989. The goal of anemia management since then has been a decrease in the proportion of patients with a less-than-target hemoglobin (Hb) level, driven by results of observational studies indicating an association between low Hb level and increased morbidity and mortality risk,1-7 as well as the established clinical benefit shown in randomized trials of transfusion avoidance8,9 and enhanced physical function.10-14 This evidence also was reflected in the 1997 Kidney Disease Outcomes Quality Initiative (KDOQI) clinical practice guidelines from the National Kidney Foundation (NKF) for anemia management in dialysis patients that recommend a Hb target range of 11-12 g/dL.15 As a result, the mean Hb level of the US dialysis
patient population increased from 9.7 g/dL in 1991 to nearly 12.0 g/dL in 2006.16 In late 2006 and 2007, multiple events took place that fundamentally changed ESA dosing practices in dialysis patients. First, new clinical trial data emerged showing either increased adverse events or no benefits in targeting higher Hb levels. The CHOIR (Correction of Hemoglobin
From the 1Health Sciences Center, University of Colorado, Denver, CO; and 2Department of Health Economics, Amgen Inc, Thousand Oaks, CA. Received May 22, 2009. Accepted in revised form September 24, 2009. Originally published online as doi:10.1053/j. ajkd.2009.09.024 on November 23, 2009. Address correspondence to David M. Spiegel, MD, 4545 E 9th Ave, Ste 160, Denver, CO 80220. E-mail: david. [email protected] © 2009 by the National Kidney Foundation, Inc. 0272-6386/09/5501-0016$36.00/0 doi:10.1053/j.ajkd.2009.09.024
American Journal of Kidney Diseases, Vol 55, No 1 (January), 2010: pp 113-120
113
114
Spiegel et al
and Outcomes in Renal Insufficiency)17 and CREATE (Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta)18 studies used ESAs to increase Hb levels in patients with chronic kidney disease not on dialysis therapy. In the CHOIR study, patients targeted to a high Hb level (13.5 g/dL) had statistically significantly more composite end point events than those targeted to lower Hb levels (11.3 g/dL). In CREATE, no benefits were found in targeting patients to a Hb level of 13.0-15.0 g/dL compared with patients targeted to a Hb level of 10.5-11.5 g/dL. Both of these studies supported what had been shown previously in the Normal Hematocrit Cardiac Trial (NHCT).19 As a result of these clinical trial data, the FDA issued a boxed warning and other product labeling changes for ESA treatment in patients with chronic kidney disease (both on and not on dialysis therapy) in March 2007. The August 2008 update of the product label recommended a Hb level range of 10-12 g/dL and included specific dose adjustment guidelines to achieve these targets, as well as guidance for dosing in hyporesponsive patients.20,21 In addition, the NKF KDOQI panel convened in March 2007 to consider findings from the new studies and issued a practice guideline update that recommended a target Hb level of 11-12 g/dL, rather than a target ⱖ11 g/dL.22 Finally, in July 2007, the Centers for Medicare & Medicaid Services (CMS) communicated changes to its existing ESA monitoring policy. In addition to encouraging downward titration in response to high Hb levels, the revised policy, effective as of January 1, 2008, decreased the maximum monthly epoetin alfa and darbepoetin alfa doses that would be reimbursed to 400,000 U/mo and 1,200 g/mo, respectively. The revised policy also decreased ESA coverage by 50% if Hb levels were ⬎13 g/dL during the current month and prior 2 months.23 The purpose of this study is to assess Hb level distribution characteristics of the dialysis patient population in the United States during the time frame when these events occurred.
METHODS A retrospective observational database was constructed from electronic patient-level data submitted by a large number (⬃4,800) of dialysis centers in the United States. These
data were obtained and purchased as part of product purchase and sales agreements maintained between Amgen Inc and its customers. Dialysis units transferred the data electronically to Thomson Reuters Healthcare Inc, which took steps to ensure proper data integration, quality assurance, and compliance with the federal Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule. The data set then was certified by an outside consultant as statistically deidentified under the HIPAA privacy rule for use within Amgen Inc; therefore, no HIPAA waiver was sought or required. The corresponding author (D.M.S.) was not privy to individually identifiable patient-level data (reviewing instead composite data analyses and graphs) and therefore no institutional review board approval was sought for his work on the data set. We classified dialysis facilities as large dialysis organizations (defined as a chain ⬎500 affiliated units), small dialysis organizations, and hospital-based dialysis centers. Of nearly 5,500 dialysis centers currently active in the United States, ⬃60% are large dialysis organizations, 30% are small dialysis organizations, and 10% are hospital-based dialysis centers. The present study database represents 100% of facilities from the large dialysis organization segment (⬃3,300) and 90% of facilities from the small dialysis organization segment (⬃1,500). Overall, nearly 87% of dialysis centers in the United States (⬃4,800) are represented. The primary reason for exclusion of facilities (ie, 100% of hospital-based dialysis centers and 10% of small dialysis organizations) was practical limitations in accessing their data because of logistical or technical hurdles. Dialysis centers that are captured in the database submitted data for all patients receiving care within that unit, and no exclusion criteria were applied. Thus, patients with all modality types, including hemodialysis and peritoneal dialysis, and patients with all coverage types, including commercial and Medicare, were represented. The database contains laboratory measurements, including all Hb measurements and intravenous medications administered by dialysis centers. Although laboratory measurements were available for all patients in the database, intravenous medications were available for only patients in large dialysis organization dialysis facilities. The primary variables used in the present study were Hb measurements. All available Hb measurements for each patient (including multiple measurements per month) were used in the analysis. The study time frame was June 2006-November 2008. This time frame was selected because November 2008 was the latest month when data were extracted for this study, and a 30 month history was judged to be sufficient for qualitative and quantitative assessment of changes due to events following publication of the CHOIR17 and CREATE18 studies in late 2006. In addition, June 2006 was judged to be a stable “baseline” point relative to which future changes can be assessed because anemia management practice patterns during this period were relatively stable and uninfluenced by future ESA safety-related events.16 The following sets of variables were calculated for each month during the study time frame for the cross-section of active patients: (1) mean Hb level; (2) standard deviation (SD) of Hb distribution; (3) skewness of Hb distribution; (4) continuous Hb distribution curve covering 0.1 g/dL intervals; (5)
Hemoglobin Distribution Changes in Dialysis percentage of Hb measurements in the following 3 ranges: ⬍10, 10-12, and ⬎12 g/dL; and (6) percentage of patients with Hb level ⬎13 g/dL for ⱖ3 months. All these calculations, except for item 6, were based on all available patient-level Hb measurements during a given month; multiple Hb values in a month were not averaged. This procedure was adopted because we were interested in describing the raw Hb distribution curve that represented total variation in the data, including within- and betweenpatient variation. Although within-patient variation during a single month was very small, averaging would have masked the information represented in this source of variation and the marginal contribution from outliers. We also performed calculations in which patient-level Hb values were averaged during a month, and the results appeared to be nearly identical to those reported in this study. Skewness calculation in item 3 represents a measure of the departure from normality for the Hb distribution curve. Skewness is zero for a normally distributed curve and nonzero if the distribution has a “tapering” characteristic toward either end. Positive skewness indicates tapering to the right, whereas negative skewness denotes tapering toward the left. The evaluation of changes in skewness can inform how the reduction in the width of Hb distribution curve has been accomplished. Calculation of the percentage of patients with a Hb level ⬎13 g/dL for ⱖ3 months (item 6) was performed considering only the last recorded Hb value for individual patients during the month. This procedure was adopted to be consistent with the manner in which the CMS calculates this information for the purpose of compliance assessment with ESA monitoring policy. The continuous Hb distribution curve (item 4) represented a smooth approximation of Hb measurement density within 0.1 g/dL intervals. This smooth approximation was necessary to avoid unnecessary visual attention to density fluctuations driven by either low prevalence buckets or rounding of reported Hb measurements to
Figure 1. Time trend of mean hemoglobin (Hb) levels. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
115 the nearest decimal by some laboratories and was obtained using PROC TPSPLINE in SAS, (SAS Institute Inc, www.sas. com) a standard smoothing tool for this purpose (comparisons with unsmoothed data also confirmed the appropriateness of this method). All analyses were conducted using SAS, version 9.1.3. P values and confidence intervals for changes discussed in results are not reported because the small P values and narrow confidence intervals (the result of a very large patient number) were not informative.
RESULTS Approximately 743,000 unique patients were available in the analysis data set during the study time frame covering June 2006-November 2008 (study period prevalence). Mean follow-up for these patients during the study time frame was 12.3 months (5th-95th percentiles, 0.5-29.5). The cohort of patients alive during any month represented the sum of those who survived from the previous month and those who represented new patients added during the month. The mean number of unique patients per month (ie, the monthly prevalence) was ⬃309,000. These patients represented ⬃734,000 unique Hb measurements each month. Mean Hb measurement frequency per patient per month was ⬃2.4 ⫾ 3.5 (SD). Mean population Hb level decreased from 12.08 ⫾ 1.48 g/dL in June 2006 to 11.71 ⫾ 1.35 g/dL in November 2008, a decrease of 0.37 g/dL (Fig 1). The periods of sharpest decrease in mean population Hb level coincided with the anticipa-
116
Spiegel et al
Figure 2. Time trend of the standard deviation (SD) of hemoglobin (Hb) distribution. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
tion and occurrence of the following events: revision of the ESA label by the FDA in March 2007, updates to the NKF KDOQI clinical practice guidelines in March 2007, and implementation of the revised ESA monitoring policy by the CMS in January 2008. The highest mean population Hb level during the study time frame was 12.14 ⫾ 1.50 g/dL in January 2007, whereas the lowest mean population Hb level was 11.71 ⫾ 1.35 g/dL in November 2008, a difference of 0.44 g/dL. In a typical cross-sectional sample of dialysis patients at any point in time, Hb values cover a wide range around the mean, and the resulting distribution is approximately normal.24 The SD has been used as a measure of the width of the Hb distribution curve. SD was calculated for each month to assess potential changes in the width of the Hb distribution curve over time. SD decreased from 1.48 g/dL in June 2006 to 1.35 g/dL in November 2008 (decrease of 0.14 g/dL; Fig 2). Figure 3 shows the skewness of the Hb distribution curve over time. The Hb distribution curve showed a departure from normality in terms of skewness. A general negative skew for the Hb distribution curve indicates that the historical Hb distribution curve probably has been skewed toward the left, indicating that significantly low Hb values relative to the peak were more common compared with significantly high
Hb values. As the SD of Hb distribution curve has decreased, the magnitude of negative skewness has increased simultaneously from ⫺0.11 in June 2006 to ⫺0.21 in November 2008, an absolute increase of 0.10. Together, these data indicate that the reduction in width of the Hb distribution curve has resulted primarily from migration of patients from high Hb ranges toward the middle. This observation also is readily apparent in Fig 4, which shows changes in continuous Hb distribution curve over time. A detailed examination of Fig 4 indicates that the area under the curve between 10 and 12 g/dL has systematically increased, primarily because of a net patient migration from high Hb ranges. In addition, the entire distribution curve appears to have shifted toward lower Hb ranges. Figure 5 provides a more granular look at the dynamics in specific Hb ranges. The percentage of Hb measurements in the ⬎12 g/dL range decreased from 51.3% in June 2006 to 40.1% in November 2008 (a decrease of 11.3 percentage points). However, the percentage of Hb measurements in the FDA-recommended target range of 10 12 g/dL increased from 41.3% in June 2006 to 50.6% in November 2008 (an increase of 9.4 percentage points), and the percentage of Hb measurements in the ⬍10 g/dL range also increased by 1.9 percentage points from 7.4% in June 2006 to 9.3% in November 2008.
Hemoglobin Distribution Changes in Dialysis
117
Figure 3. Time trend of the skewness of hemoglobin (Hb) distribution. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
Figure 6 shows that the proportion of patients with Hb levels persistently ⬎13 g/dL for ⱖ3 months decreased from 5.1% in June 2006 to 2.2% in November 2008 (a decrease of 2.9 percentage points). ESA-dosing behavior was examined for patients with a Hb level persistently ⬎13 g/dL for at least 3 months in November 2008 who also received care in large dialysis organizations. Approximately 64% of these patients received zero ESA dose during November 2008. That is, these patients naturally remained
in the Hb ⬎13 g/dL range without requiring further administration of an ESA.
DISCUSSION Changes implemented by the FDA and CMS in light of safety concerns regarding ESA treatment in dialysis patients have had a profound impact on anemia management practice patterns in the United States. A population-level Hb distribution curve that would be aligned best with the intent of these changes would maximize the
Figure 4. Continuous hemoglobin (Hb) distribution curve over time.
118
Spiegel et al
Figure 5. Percentage of hemoglobin (Hb) measurements in specific ranges. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
number of patients in the target range of 10-12 g/dL, minimize the number of patients in the ⬍10 and ⬎12 g/dL ranges, and minimize the number of patients who are serially exposed to ⬎13 g/dL. To accomplish the desired result of maximizing patients in the target range, one could affect the mean population Hb level, SD, or skewness characteristics of the Hb distribution curve. The Hb distribution curve described in the present study represents a “snapshot,” or a crosssectional point-in-time view. However, when in-
Figure 6. Percentage of patients with hemoglobin (Hb) ⬎13 g/dL for ⱖ3 months. Abbreviations: CHOIR, Correction of Hemoglobin and Outcomes in Renal Insufficiency; CREATE, Cardiovascular Risk Reduction by Early Anemia Treatment With Epoetin Beta; EMP, Erythropoiesis-stimulating agent Monitoring Program; FDA, Food and Drug Administration; NKF KDOQI, National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative.
dividual patients are followed up over time, they are seen to oscillate between low, intermediate, and high Hb ranges, and extremely few patients are successfully managed in a narrow range over time.24 It is not clear whether these oscillations represent an intrinsic clinical characteristic of patients, transient intercurrent events, or are influenced by anemia management practice patterns. The nature of the snapshot Hb distribution curve thus is influenced largely by the nature of the patient migration pattern between low, intermedi-
Hemoglobin Distribution Changes in Dialysis
ate, and high Hb ranges. The information regarding changes in mean, SD, and skewness of the Hb distribution curve from this study can be used to generate useful hypotheses about the types of patient migration patterns that are consistent with these changes. Although the decrease in mean population Hb level presumably represents decreased ESA dosing and lower point targets, both the decrease in the SD and increase in the negative skewness could be explained by a preferential patient migration from high Hb ranges toward the target range, accompanied by some patient migration from target range to below target. Although we do not yet understand the exact changes in the nature of ESA dosing algorithms that might have led to the observed changes in the SD and skewness of the Hb distribution curve, it is an area worthy of further investigation. One possible explanation for the observed decrease in the SD of the Hb distribution curve, as shown in Fig 2, may be changes in specific Hb thresholds that trigger a downward dose adjustment, as well as changes in the magnitude and frequency of these downward dose adjustments in response to the factors described in this study. Other factors, such as changes and/or integration of iron and ESA dosing protocols, also may have contributed to the observed changes in the Hb distribution curve.25 Based on current guidelines, ESA dosing protocols should strive to maximize the number of patients with Hb levels in the 10-12 g/dL range while minimizing the proportion of patients in the high-Hb range. However, this should not be at the expense of an increase in the proportion of patients in the low-Hb range. To achieve this level of control, anemia management practice patterns that allow for timely changes in Hb levels through more frequent Hb monitoring and more frequent dose adjustments may be preferred over anemia management practice patterns based on infrequent Hb measurements requiring potentially large dose changes. Additional factors, such as the nature of ESA dose-holding practices, also may influence the nature of Hb distribution. For example, a recent study evaluating individual patient-level data in a single center suggested that a principal cause of Hb level decreases to ⬍11 g/dL was withholding of ESAs for a Hb level ⬎13 g/dL.26 However, not all published literature support this conclusion.27 If
119
ESA doses are withheld to manage high Hb level excursions, it seems prudent to monitor Hb levels frequently so that treatment is reinitiated before Hb levels decrease too much. More studies evaluating population-level data are needed to understand if a more optimal approach toward ESA dosing algorithms can be developed. Some important limitations of our study need to be considered. Although the sample size was relatively large, incorporating data from nearly 87% of dialysis centers in the United States, patients in the remaining 13% of dialysis centers may represent a systemic bias because excluded facilities represented 100% of the hospital-based dialysis center facilities and 10% of the small dialysis organization facilities. Although the excluded hospital-based dialysis center segment could have different patient and anemia management characteristics from the facilities studied, the 10% excluded small dialysis organization facilities also could have characteristics that are different compared with the rest of the small dialysis organization segment because the excluded small dialysis organization facilities were primarily those that did not have a fully functional electronic system for patient-level data collection. Another limitation of our study was that we did not have demographic or comorbidity information for patients in the analysis database and thus were unable to summarize patient clinical characteristics. However, for the facility segments under consideration in this study, these characteristics have been well documented and reported in several other sources.5,6,16,24 In addition, we consider it to be extremely unlikely for potential changes in patient characteristics to be an explanation for the dynamics reported in this study because patient characteristics change over a much longer time horizon and anemia management practices have a far greater and immediate impact on the nature of Hb level distribution in this patient population. Another limitation of the study is that although Hb data from outpatient dialysis sessions have been represented, data from inpatient hospitalization records were not included. However, because only ⬃4% of dialysis sessions per patient-year occur in the inpatient setting,16 any potential changes in practice patterns during hospitalizations are very unlikely to explain the population-level results discussed in this study.
120
Spiegel et al
Despite these limitations, this study offers important findings regarding changes in Hb level distribution characteristics for a large number of patients. How these changes will ultimately affect patient outcomes in the longer term is unclear and is a topic worthy of further investigation.
ACKNOWLEDGEMENTS The authors thank Santosh Sastry, MS, for help with data access and SAS programming and Yeshi Mikyas, PhD, for editorial support. Support: Information on funding sources is listed in the financial disclosure. Financial Disclosure: This study was supported by Amgen Inc, which markets ESAs. Dr Spiegel has received consulting income and grant support from Amgen Inc and has served on advisory boards and speakers bureaus for Amgen Inc. Drs Khan, Krishnan, and Mayne are employees of Amgen Inc.
REFERENCES 1. Collins AJ, Li S, St. Peter W, et al. Death, hospitalization, and economic associations among incident hemodialysis patients with hematocrit values of 36 to 39%. J Am Soc Nephrol. 2001;12:2465-2473. 2. Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, Barre PE. The impact of anemia on cardiomyopathy, morbidity, and mortality in end-stage renal disease. Am J Kidney Dis. 1996;28:53-61. 3. Lowrie EG, Huang WH, Lew NL, Liu Y. The relative contribution of measured variables to death risk among hemodialysis patients. In: Hingman FE, ed. Death on Hemodialysis: Preventable or Inevitable. Boston, MA: Kluwer Academic Publishers; 1994:121-141. 4. Ma JZ, Ebben J, Xia H, Collins AJ. Hematocrit level and associated mortality in hemodialysis patients. J Am Soc Nephrol. 1999;10:610-619. 5. Ofsthun N, Labrecque J, Lacson E, Keen M, Lazarus JM. The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients. Kidney Int. 2003;63:1908-1914. 6. Regidor DL, Kopple JD, Kovesdy CP, et al. Associations between changes in hemoglobin and administered erythropoiesis-stimulating agent and survival in hemodialysis patients. J Am Soc Nephrol. 2006;17:1181-1191. 7. Roberts TL, Foley RN, Weinhandl ED, Gilbertson DT, Collins AJ. Anaemia and mortality in haemodialysis patients: interaction of propensity score for predicted anaemia and actual haemoglobin levels. Nephrol Dial Transplant. 2006;21:1652-1662. 8. Eschbach JW, Abdulhadi MH, Browne JK, et al. Recombinant human erythropoietin in anemic patients with end-stage renal disease. Results of a phase III multicenter clinical trial. Ann Intern Med. 1989;111:992-1000. 9. Eschbach JW, Egrie JC, Downing MR, Browne JK, Adamson JW. Correction of the anemia of end-stage renal disease with recombinant human erythropoietin. Results of a combined phase I and II clinical trial. N Engl J Med. 1987;316:73-78.
10. Canadian Erythropoietin Study Group. Association between recombinant human erythropoietin and quality of life and exercise capacity of patients receiving haemodialysis. BMJ. 1990;300:573-578. 11. Beusterien KM, Nissenson AR, Port FK, Kelly M, Steinwald B, Ware JE Jr. The effects of recombinant human erythropoietin on functional health and well-being in chronic dialysis patients. J Am Soc Nephrol. 1996;7:763-773. 12. Evans RW, Manninen DL, Garrison LP Jr, et al. The quality of life of patients with end-stage renal disease. N Engl J Med. 1985;312:553-559. 13. Evans RW, Rader B, Manninen DL. The quality of life of hemodialysis recipients treated with recombinant human erythropoietin. Cooperative Multicenter EPO Clinical Trial Group. JAMA. 1990;263:825-830. 14. Laupacis A. Changes in quality of life and functional capacity in hemodialysis patients treated with recombinant human erythropoietin. The Canadian Erythropoietin Study Group. Semin Nephrol. 1990;10:11-19. 15. National Kidney Foundation. DOQI Clinical Practice Guidelines for the Treatment of Anemia of Chronic Renal Failure. Am J Kidney Dis. 1997;30(suppl 3):S192-240. 16. US Renal Data System. USRDS 2008 Annual Data Report. Bethesda, MD: The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2008. 17. Singh AK, Szczech L, Tang KL, et al. Correction of anemia with epoetin alfa in chronic kidney disease. N Engl J Med. 2006;355:2085-2098. 18. Drueke TB, Locatelli F, Clyne N, et al. Normalization of hemoglobin level in patients with chronic kidney disease and anemia. N Engl J Med. 2006;355:2071-2084. 19. Besarab A, Bolton WK, Browne JK, et al. The effects of normal as compared with low hematocrit values in patients with cardiac disease who are receiving hemodialysis and epoetin. N Engl J Med. 1998;339:584-590. 20. Epogen (epoetin alfa) [prescribing information]. Thousand Oaks, CA: Amgen Inc; 2008. 21. Aranesp (darbepoetin alfa) [prescribing information]. Thousand Oaks, CA: Amgen Inc; 2008. 22. National Kidney Foundation. KDOQI Clinical Practice Guideline and Clinical Practice Recommendations for Anemia in Chronic Kidney Disease: 2007 update of hemoglobin target. Am J Kidney Dis. 2007;50:471-530. 23. Centers for Medicare & Medicaid Services. CMS Manual System. Pub 100-04 Medicare Claims Processing. Transmittal 1307. http://www.cms.hhs.gov/Transmittals/ Downloads/R1307CP.pdf. Accessed January 15, 2009. 24. Lacson E Jr, Ofsthun N, Lazarus JM. Effect of variability in anemia management on hemoglobin outcomes in ESRD. Am J Kidney Dis. 2003;41:111-124. 25. Chan K, Moran J, Hlatky M, Lafayette R. Protocol adherence and the ability to achieve target haemoglobin levels in haemodialysis patients. Nephrol Dial Transplant. 2009;24:1956-1962. 26. Spiegel DM, Gitlin M, Mayne T. Factors affecting anemia management in hemodialysis patients: a singlecenter experience. Hemodial Int. 2008;12:336-341. 27. Weiner DE, Miskulin DC, Seefeld K, et al. Reducing versus discontinuing erythropoietin at high hemoglobin levels. J Am Soc Nephrol. 2007;18:3184-3191.